International workshop on computer aided processing of intertextuality in ancient languages
Compte-rendu de cet atelier : C. Crosnier, L. Mellerin, « International Workshop on Computer Aided Processing of Intertextuality in Ancient Languages (2-4 juin 2014, Lyon) : bilan et perspectives », Bolletino di Studi Latini XLIV (2014), f. II, p. 255-260.
Bâtiment Blaise PASCAL, INSA, Campus de la Doua, Villeurbanne (France).
All the sessions will take place in room 501.337, on the 3rd floor.
coorganized by HiSoMA (UMR 5189, Lyon), LIRIS (UMR 5205, Villeurbanne) and by the Göttingen Centre for Digital Humanities (e-TRAP), with the support of the National Research Agency (ANR Biblindex) and the Partner University Fund (PUF).
This workshop was initiated as the conclusive meeting of the project Biblindex, funded by the French National Research Agency (ANR), which aims at establishing an exhaustive statement of the biblical references found in the texts of the Late Antiquity and the Middle Ages. At this meeting will be gathered computer scientists and digital humanists, specialists of corpora written in ancient languages. The planned sessions aim to present the state of art regarding concepts and technics used to process quotations in ancient languages. A lot of projects work nowadays on various corpora, asking similar questions about text-reuse. Comparing experiments, we hope to clear perspectives to mutualize developments and methodological choices, in order to build a federative project at the European scale in the coming years.
The first session will be devoted to mutual project presentations. Afterwards, the various stages of quotation processing will be discussed in four workshops. The first two of them will tackle the preparation of sources and the automatic retrieval of concording text places: interests and complementarities of statistical and linguistic approaches will be compared. The next two will focus on the conceptual definitions, the modelling of the unstable idea of “quotation” and the XML-TEI encoding to implement for its characterization, in close interdependence with visualization choices.
Registration needed: email@example.com
Monday, 2nd June
Session 1: projects overview
This session aims to inform the participants about some bigger projects being available. Presentations of the projects give an overview to the research questions, used data, and project objectives.
Biblical Quotations in Patristic Texts
Moderator: Yasmine Ech Chael (HiSoMA, Lyon)
The Biblindex project, funded by the French National Research Agency and supported by the Institut des Sources Chrétiennes in Lyon, helped initiate the creation of an index of biblical quotations and allusions in Early Christian Literature, which plans to become exhaustive. It seeks to link a corpus of biblical texts – collections of scriptural books which were originally written in various languages and translated early in their history – with a corpus of ancient and medieval authors, who refer to the Bible as a fixed entity yet at the same time contribute through their quotations to the form and concept of ‘the text’.
The first step of our enterprise was to establish a nomenclature for our sources, e.g. providing keys to precise lists of authors and works, instituting fixed points of reference and the identification of reliable critical editions and other types of relevant material. Then biblical text re-uses have been gathered. At present, BiblIndex consists of a comprehensive inventory of 700.000 biblical quotations and allusions in Early Greek and Latin Christian Literature. The database now includes eleven biblical texts written in different ancient and modern languages. A multilingual concordance between those Bibles has been created, allowing the user to visualize the biblical text in order to compare it with the quotations found. We already propose simple search forms to access to the data on the website http://www.biblindex.org.
Each entry on the website offers a series of numbers indicating the chapter and verse of the biblical text, its location in the patristic writing and the corresponding page and line numbers in the reference edition. We now need to make links between the BiblIndex database and the existing textual databases.
9:30 The Digital Greek Patristic Catena (Athanasios Paparnakis, Aristotle University, Thessaloniki)
Digital Greek Patristic Catena: Based on a collection of nearly 350.000 biblical quotations in the Patristic texts from Migne’s Patrologia Graeca edition of nearly 900 ecclesiastical authors and 6000 works, a database has been developed following the methodology of the ancient hermeneutical tool of the catenae. The database utilises four sets of information:
a) the Greek text of the Bible and modern translations,
b) the authors and works,
c) the patristic text in two forms: short patristic passages containing the biblical quotations and an image of the full page and
d) indexes of subjects, names and contents.
Early Christian Latin writers frequently cite, adapt and discuss the text of the Bible; since they are using early forms of the biblical text, their evidence can provide important information about that text as well as offering insight into how the Bible was read and interpreted in the early period of its history.
The Vetus Latina Iohannes Project is producing an edition of the Old Latin materials for the Gospel of John that provide evidence for early alternatives and predecessors to today’s familiar Vulgate Latin translation; two fascicles have been published so far covering John 1-9. There are two sorts of materials: manuscripts with elements of pre-Vulgate textual traditions and citations from patristic authors. The project thus has two technical components: an electronic edition of the manuscripts and a database of citations of the Gospel of John in patristic works. Data from both of these is brought together in the final printed edition.
The project has made electronic transcriptions of the Gospel manuscripts containing texts with a potential Old Latin element following the TEI guidelines with specific encoding of features found in biblical manuscripts. The online edition of these manuscripts (http://www.iohannes.com/vetuslatina) allows access to both transcriptions of individual manuscripts and a synopsis of each verse in all the manuscripts. In the case of the Gospel of John in the Old Latin Bible, the main strands of Old Latin text are represented by extant manuscripts and the transcriptions of these manuscripts thus provide a framework to add the patristic citations.
For the past century details of patristic citations and references have been manually collected from printed editions on index cards by the Vetus Latina Institute at Beuron; these Beuron cards, some typed, some handwritten, have now been digitally imaged. The Vetus Latina Iohannes Project has transcribed around 60,000 citations for the Gospel of John from these cards and other sources into Excel spreadsheets which have now been transformed into an online database with related details of authors, works and editions. Corralling the diverse and sometimes awkward material into a useable database, while maintaining compatibility with legacy systems of referencing, is not without its challenges. From the material in the database, the edition presents citations of each verse and an index of readings for each verse.
The COMPAUL project is investigating the earliest commentaries on the Pauline Epistles as sources for the biblical text. There are an exceptional number of surviving commentaries on the Pauline Letters, in Greek and especially in Latin, composed in the 4th Century and including works by significant figures of the period such as Augustine, Jerome and Pelagius. These provide significant evidence for pre-Vulgate textual traditions earlier than those represented in surviving manuscripts. This project thus provides a valuable preliminary to future editions of the Pauline Epistles.
The project has been analyzing the biblical text in each commentary using electronic transcriptions of commentary manuscripts and biblical manuscripts to identify and explore each author’s text of the epistles. This involves in particular comparing the biblical text found in the lemma of the commentaries with the references to the biblical text in the authorial exegesis since later generations may have replaced or adapted the biblical text, especially in the lemma, to forms current in their time. This data is being collected in an extension to the patristic database developed for the Vetus Latina Iohannes project. We are also starting to explore whether characteristic readings in the biblical text can be used to trace later authors using a commentary rather than a biblical manuscript.
The ECM-project aims for a new reconstruction of the initial text of the Greek New Testament and the critical edition of the Acts of Apostles as part of this project is currently nearing completion. This project overview will cover a searchable online database for patristic citations established explicitly for the work on this edition as well as a user interface that allows relating the gathered patristic data with the manuscript data prepared for the critical apparatus.
Text Re-use in the Digital Humanities
The presentation gives an overview of the recently planned activities of the eTRACES project and the eTRAP Early Career Research Group. The starting point is a requirements analysis for the mining and retrieval tool. Furthermore, the presentation introduces the TRACER framework. This is followed by a brief overview of the use case scenarios.
The goal of this paper is to present a project of the Humboldt Chair of Digital Humanities at the University of Leipzig: the Leipzig Open Fragmentary Texts Series (LOFTS). LOFTS establishes open editions of ancient works that survive only through quotations and text re-uses in later texts (i.e., those pieces of information that humanists call “fragments”). LOFTS has two goals: 1) digitize paper editions of fragmentary works and link them to source texts; 2) produce born-digital editions of fragmentary works. LOFTS uses both XML and RDF, the CTS/CITE Architecture, and different data models like the PROV-O ontology, the Systematic Assertion Model (SAM), and the Open Annotation (OA) Core data model.
LOFTS has three submain projects: 1) the Digital Fragmenta Historicorum Graecorum (DFHG); 2) the Digital Marmor Parium (DMP); 3) the Digital Athenaeus.
The Sharing Ancient Wisdoms project, funded by HERA, ran from 2010 to 2013. Our aim was to explore ways in which to present and analyse citations; we focused on the later Greek tradition, and translations into Arabic. We published texts in TEI compliant XML; we developed CTS identifiers; and we started building an ontology to express the various kinds of relationships with which we were dealing. Above all, we aimed to connect citations – both in collections, such as the gnomologia edited by Denis Searby and the Swedish team) or embedded in continuous texts (Kekaumenos, edited by Charlotte Roueché) with one another, and with their source texts. Elvira Wakelnig, and the Vienna team, further explored the relationships of Greek to Arabic collections and Wisdom Literature, and the further journeys of such materials into Latin and Spanish.
This was only a first attempt; but we have made all our materials, and most importantly, our ontology, available, for others to use. This paper will present our results; we hope to discover other colleagues who can make use of our work, and take it forward.
The project “Corpus der arabischen und syrischen Gnomologien” (CASG) was funded by the Fritz Thyssen Stiftung (2010-2012) and is supervised by Dr. Ute Pietruschka, who together with her assistants will set up a database and digital repository containing all available collections of Gnomologia written in Arabic and Syriac.
In the field of literature, Late Antiquity marks a shift in literary style with a preference for encyclopedic works, consisting of summaries of earlier works. Collections of sayings like the Gnomologium Byzantinum, the Gnomologium Parisinum and the Christian collection of John of Damascus were a popular kind of literature and represented an important source for the transmission of knowledge. Their popularity can be seen as connected to the compilation of anthologies and handbooks on varying topics that were intended for a quick orientation in a certain field of knowledge. A gnomologium is a collection of short sentences and anecdotes attributed to more or less commonly known philosophers, poets, rulers, politicians. These collections sometimes are arranged by alphabetical order of the authors, some are arranged by topics while others seem to have no internal order at all. Most gnomologia are no pure gnomologia as such: amongst real gnomai (sentences), apophtegms, chreiai or diatribes can be found. The gnomological tradition was not limited to Byzantine literature – on the contrary it can be found in all Mediterranean civilizations and those under their cultural influence. We thus find Greek gnomological material even in Coptic and Ethiopic manuscripts. The Arabic gnomologia are the largest extant group of collections beside the Greek ones. The project aims at illuminating the transmission of Syriac and Arabic gnomologia, which have been studied even less than the Greek collections and at comparing them regarding their different aspects.
Session 2. Natural Language Processing Methods for Retrieving Texts and Computing Text Similarities
14:30 Introduction: overview of the possible methodologies used in text retrieval (Marco Büchler, GCDH)
With the emerging amount of available digital data, there is a need for more automatic methods. This session introduces work that is related to NLP methods that are relevant in the case of Big Data. In this respect, the computational complexities are highlighted and discussed. The presentation ends with a list and explanation of some NLP problems connected with Big Data.
This work focuses on refining and finding quotations of Greek Church Fathers within the Greek versions of the Tanakh and the New Testament. Our corpus features over 700 works of various Church Fathers, which amounts to around ten millions words.
In order to reach our aim, we had to explore the notion of quotation as relevant to the first centuries of our era, and to discuss the efficiency and usability of modern digital approaches, including their evaluation metrics. We mainly studied unsupervised approaches, either statistical, statistico-structural and statistico-semantic. We ended up building a generic and flexible solution based on a robust document model and an algorithm merging several other approaches. We are currently working on implementing this solution within the GATE framework to facilitate reusability.
In this presentation, we will talk about aspects of the challenges, the model we built to ensure genericity and flexibility, the algorithm we built from the various approaches we explored and discuss our results applying this algorithm to two different control sets.
15:15 QuotationFinder – Searching for Quotations and Allusions in Greek and Latin Texts (Luc Herren, London)
Traditional search functions of software used to access the TLG (Thesaurus Linguae Graecae) or the CLCLT (CETEDOC Library of Christian Latin Texts) are not well suited for finding quotations and allusions. A search with a Boolean « or » would yield too many matches, as most texts contain at least one of the common words in our search string. A search with a Boolean « and, » however, would yield too few, as ancient authors often left out words when quoting without us having a chance to know in advance which ones they would keep, resulting in matches being missed because they lack just one (possibly dispensable) word in our search text.
QuotationFinder gets around this problem by using more sophisticated criteria for determining if a given text is a quotation/allusion. It reads text files exported from the TLG or the CLCLT and considers 5 parameters when it encounters words from your search text: It looks at the number of words matched within a reasonable number of lines; whether the exact form of a word is matched, or a different form of the same word, or a form of a cognate; how close the matched words are to each other; how rare the matched words are; and, finally, to what degree the sequence of the words matches your search text. QuotationFinder then produces an ordered list with the most exact quotation first and the loosest verbal parallel last.
The study of intertextuality, or how authors make artistic use of other texts in their works, has a long tradition, and has in recent years benefited from a variety of applications of digital methods. This paper describes an approach to detecting the sorts of intertexts that literary scholars have found most meaningful, as embodied in the free Tesserae website http://tesserae.caset.buffalo.edu/. Tests of Tesserae Versions 1 and 2 showed that word-level n-gram matching could recall a majority of parallels identified by scholarly commentators in a benchmark set. But these versions lacked precision, so that the meaningful parallels could only be found among long lists of those that were not meaningful. The Version 3 search described here adds a second stage scoring system that sorts found parallels by a formula accounting for word frequency and phrase density. Testing against a benchmark set of intertexts in Latin epic poetry shows that the scoring system overall succeeds in ranking parallels of greater significance more highly, allowing site users to find meaningful parallels more quickly. Users can also choose to adjust recall and precision by focusing only on results above given score levels. As a theoretical matter, these tests establish that lemma identity, word frequency, and phrase density are important constituents of what make a phrase parallel a meaningful intertext.
16:30 Round-table discussion: methodological discussion about statistical approaches
Leader: Marco Büchler
Processing natural language means dealing with the complexity of human interaction. This includes language evolution, the many different dialects, but also any kind of error or change in purpose of meaning. This complexity brings a need for normalization. What is a good level of normalization? If the text is kept to an under-normalized level, then relevant text re-use remains undiscovered. If the text is over-normalized, it becomes easy to spot irrelevant noise. Together with this topic of discussion, the round-table shares experiences and lessons learned.
Tuesday, 3rd June
Session 3: linguistic approaches / language specificities
This section aims to introduce relevant linguistic approaches depending on languages themselves.
Since the 1970s, the Eep Talstra Centre for Bible and Computer (ETCBC; formerly known as the Werkgroep Informatica Vrije Unviversiteit) has invested in building a database of the Hebrew Bible incorporating linguistic information at the level of words, phrases, clauses and clause relations. Since 1999, when the ETCBC joined forces with the Peshitta Institute Leiden, the analysis of Syriac texts came into focus and two successful research projects funded by the Netherlands Organization for Scientific Research dealt with multilingual comparison of Hebrew, Syriac and Jewish Aramaic Bible texts: CALAP (1999-2005) and Turgama (2005-2010).
Whereas the ETCBC started as the work of lonely pioneers, the emergence of Digital Humanities research centers and the high position of DH on research agenda’s opens up a new potential for the computational exploration of Hebrew and Aramaic text resources, combining various methods of text comparison and text clustering based on linguistic features (comparing syntax trees; vocabulary analysis; statistical glossing) and uninformed or ‘blind’ techniques (n-gramming; Normalized Compression Distance; Kolmogorov complexity). We are now running a pilot to test these methods by training the computer to identify OT quotations in the NT. This might be a first step towards a contribution to a project that aims at recognizing biblical quotations in larger corpora.
9:45 GREgORI: Softwares, linguistic data and corpus for Ancient GREek and ORIental languages (Tamara Pataridze; Emmanuel Van Elverdinghe,Université catholique de Louvain – CIOL (Centre d’études orientales, Institut orientaliste de Louvain)
The Research Project in Greek Lexicology (UCL: Prof. B. Coulie, Dr. B.Kindt) pursues a twin goal: on the one hand, constituting an electronic dictionary of Ancient and Byzantine Greek, on the other hand, producing lemmatized concordances. The dictionary gathers linguistic data directly stemming from corpus-based observations, without any restriction regarding the handled texts’ date, literary genre, language level or dialect; moreover, every occurring word is recorded. Each corpus analysis thus provides a comprehensive lexical inventory of the processed texts, lemmatized and perfectly disambiguated. On this base are produced various lexicological tools: lemmatized concordances, lemmatized indexes, reverse indexes, frequency indexes, index of words common or specific to two corpuses or corpus parts.
The know-how developed for Greek is now being extended to the other languages of Christian Orient within the framework of the GREgORI: Softwares, linguistic data and tagged corpus for Ancient GREek and ORIental languages project, the aim being as well unilingual as multilingual. Emmanuel Van Elverdinghe studies the formulaic style found in Armenian colophons: the purpose is to navigate through a corpus spanning several centuries in order to extract stereotypical lexical and syntactic patterns, whose repeating or varying provide information concerning textual transmission as much as the copyists’ customs. Tamara Pataridze works out bilingual Greek-Georgian lemmatized lexicons of Gregory Nazianzen’s Discourses: a processing inspired by the alignment methods leads to accurate analysis of the translation techniques used by the authors of the Oriental versions of the Theologian’s Discourses.
La plateforme DataLift (http://www.datalift.org/) est le résultat d’un projet ANR-Contint qui vient de se terminer. Il s’agit d’une solution « tout en un » : bâtie sur une socle technique conçu pour une forte modularité. La métaphore de l’ascenseur (Lift) correspond au processus d’élévation qui se décompose en 5 étages. Le premier est destiné à la capture de jeux de données structurés mais hétérogènes et à la désignation d’ontologies permettant de les décrire. Le deuxième étage comprend les convertisseurs/mappeurs qui appliquent les ontologies et produisent des jeux de données RDF. Au troisième étage, ces jeux élevés (ils ne sont plus englués dans la gangue de leurs formats d’origine) sont stockés dans un triple store. Le quatrième étage est dédié à l’interconnexion de données : en mettant à profit les alignements que les ontologies mobilisées au 2e étage, on peut produire de nouvelles données. Le dernier étage est dévolue à l’exploitation des données élevées et interconnectées. Si l’on se rapporte à l’échelle proposée par Tim berners-Lee (http://www.w3.org/DesignIssues/LinkedData.html, 2), relative aux données du « web of data », DataLift permet de passer aux données 5 étoiles.
L’exploitation des données peut se faire à l’aide de requêtes exprimées en SPARQL, c’est-à-dire dans une modalité logique très expressive. Il est par exemple possible de faire de l’inférence. Une telle plateforme peut donc être utilisée dans le cadre de la linguistique si l’on ramène la problématique à des jeux de données structurées et des requêtes logiques (logique de 1er ordre).
C’est ce que nous avons fait pour une première approche du traitement de l’arménien ancien. Le développement était très léger, pas plus de quelques heures et le résultat est très encourageant (http://www.kepeklian.com/blog/2014/02/27/analyse-grammaticale-armenien-classique-datalift/, 3). La même approche peut être reproduite à l’envi pour d’autres langues. Nous avons créé un tokenniseur unicode de l’arménien classique, constitué une base explicite de lemmisation et utilisé un dictionnaire de traduction arménien / français. Nous avons donc trois jeux de données sous forme de fichier : premièrement les données produites par le tokenniseur (http://www.kepeklian.com/tokenisation_arm/tokenisation2.php, 4) à partir d’un texte brut, deuxièmement la base de lemmisation, troisièmement le dictionnaire. Les ontologies sous-jacentes sont produites « à la volée » par la plateforme lors de la capture des jeux de données. Nous montrerons quelques requêtes SPARQL qui permettent de répondre à des questions d’analyse grammaticale du texte, ou de traduction.
The Text Alignment Protocol (TAP for short) is a nascent XML data model and set of recommended best practices to serve many people who wish to encode, exchange, and study texts and their translations, quotations, paraphrases, adaptations, and summaries. TAP files are designed to be maximally readable and editable by both humans and machines, independent of (but compatible with) third-party software. The language-agnostic protocol is modular, allowing it to extend and grow, and permitting editors and researchers to work independently, collaboratively, and within their preferred assumptions and purposes. Although expressive of scholarly nuance and complexity, TAP files are meant to benefit scholars and nonscholars alike—anyone interested in the detailed study of ancient, medieval, and modern translations, paraphrases, and related textual reuse.
In this presentation I introduce the TAP data model by discussing its structure, the envisioned distributed workflow, and the four modules that have so far been developed: transcriptions (modified TEI), alignment, lexicomorphology, and grammar. I focus on the principles that have guided TAP design, and discuss challenges and successes that beset a search for a way to encode texts that synthesizes the simplicity cherished by computer scientists, the complexity treasured by philologists, and the self-awareness prized by theorists.
I summarize creative applications that TAP could serve, i.e., multilingual publishing, language learning, and semi-automated machine translation. But I also conclude with a summary of challenges that remain, not least of which is the lack of a broad range of collaborators.
11:30 Round-table discussion: the relationship between the concept of language and linguistic approaches
Leader: Sara Schulthess (University of Lausanne)
– Lexical, semantic approaches: lemmatization, standardization: to what extent is each language specific?
– Syntactic approaches: to what extent is each syntax specific?
– Articulation of different linguistic approaches
– What can be reused from one language to another? Step by step, which tools could be thought generical?
Session 4: Towards a Digital Ecosystem for Text Re-use and its Applications
Moderator: Elöd Egyed-Zsigmond (LIRIS, Lyon)
14:00 Introduction – Requirements for a Digital Ecosystem (Marco Büchler, GCDH)
A lot of text re-use has been discovered over the last centuries being re-published in books. Some of it kept existent while others got lost. In a digital world, data can be easily copied so that the information loss can be more or less ignored. What are the requirements for a digital system as this? This brief introduction gives an overview to some requirements on different levels.
14:15 TEI Encoding Patterns for Citations, Quotations and Allusions: what’s « in Stock » ? (Emmanuelle Morlock, HiSoMA, Lyon)
The concept of citation implies a decontextualization and an incorporation, with different degrees of accuracy and expliciteness. In a practical perspective, representing this incorporation with markup means aligning or combining the logics of two texts. It also requires to take pay special attention to the explicitness of the encoding and the disambiguisation of implicit information, to permit for example comparative extractions for intertextuality studies. A great deal of flexibility and a global view of what’s available in the encoding scheme is thus needed. The talk will give a synthetic view of the main mechanisms offered by TEI guidelines. The examples used will be taken from encoding samples realized in the context of the Biblindex project.
Originally designed to meet the needs of the Homer Multitext project, the CITE architecture defines URN notations for two fundamental kinds of scholarly citation: the CTS URN for textual citation and CITE Collection URN for citing other kinds of discrete objects. Objects in CITE Collections may or may not be ordered; in contrast, citable nodes of text are always ordered and situated in both a citation hierarchy and a work hierarchy.
Software for working with data in the CITE architecture includes a suite of network services for retrieving objects identified by URN, and utilities for automatically building an RDF graph of all relations specified by URN citation from an archive of CTS and CITE Collection data in simple text formats. (See http://cite-architecture.github.io/)
In this paper, we show how URN-aware applications can use such a scholarly graph to align any kind of analytical data set with textual citation so that analytical data can be identified and retrieved in terms of the order and hierarchy of textual citation. We will illustrate how citation of analytical data as a further hierarchical extension of textual citation can capture many kinds of intertextual relations, including: “fragments” and quotations of texts of one text by another; direct commentary by one text on another; shared physical and paleographic features of different texts; relations of features of diagrams to accompanying texts; and comparable geographic, temporal and quantitative contents of different texts.
15:00 Declaring Quotations through Canonical Reference Numbers in the Text Alignment Protocol (Joel Kalvesmaki, Dumbarton Oaks)
In this presentation, I show how the Text Alignment Protocol (TAP) may be used to encode quotations (verbatim or not) that one work makes of another. I use as a simple example Evagrius Ponticus’s Practicus, a fourth-century Greek text about monastic ascesis. The short preface, my focus here, quotes five times from the Scriptures, including once from the Psalter, and so illustrates the challenge of how to encode a quotation from a work that has alternative canonical numbering schemes.
I survey how the challenge can be tackled natively in standard Text Encoding Initiative (TEI) files. I argue that key problems beset the standard TEI model: ambiguity, siloed (single-project) terminology, and lack of nuance. I then show how the issue is handled in TAP files: the quoted work (the target) and the quoting work (the source) are transcribed in separate TEI files, customized to be restricted to a single version of a work, structured according to a single canonical reference system that has a typology. Between the source and target stands a simple alignment file that declares the distance between the two, the types of textual reuse (in this case quotation), and the corresponding canonical references between the target and the source. I illustrate ways that doubt, ambiguity, and changing text reuse strategies can be declared in a TAP alignment file. Using the example of the Psalter, I also show how TAP can be used to reconcile conflicting canonical reference systems for the same work.
The workshop will be devoted to the exploration of the concept of “quotation” and to all the digital processing arising from the definitions. Intertextuality, reuse, different degrees of duplication (literal quote, explicit quote, more or less involuntary quote), different degrees of accuracy, of imitation, or pastiche : all the materials which support this survey will be texts, and for the most, ancient texts.
It could be interesting to show that quite the same mechanisms are at work in the use of images and that the typology we try to build for text-materials may be transposed to images. Furthermore, sometimes, both textual and visual citation are used together, the two ways of quoting reinforcing each other.
How can we use digital tools — and which tools ? — to search for and to analyze visual quotation ?
Some examples taken from the Collection of Greek inscriptions of the Louvre Museum will support the analysis.
DFHG is producing a digital edition of the five volumes of Karl Müller’s Fragmenta Historicorum Graecorum (FHG) (1841-1870) according to the EpiDoc Guidelines and the CTS/CITE Architecture. This project has produced a catalog of more than 600 fragmentary authors edited by Müller, which is part of the Perseus Catalog, and a set of Guidelines that are contributing to the Epidoc community.
DMP is working on a digital edition of the so called Marmor Parium, a Greek universal chronicle on a marble stone.
16:45 Round-table discussion: Is a common methodology possible?
Leader: Greta Franzini (Univ. Leipzig)
There are several ways in which to store text re-use data, none of which are considered the standard. However, at this stage, in order to come up with one global Digital Ecosystem, it has become necessary to set a common standard.
The round-table will, for example, discuss the pros and cons of in-document vs. stand-off markup. In addition, there will be questions to address, such as, what would the digital ecosystem look like: one local repository, or better a distributed infrastructure? What are good common standards for the communication between existing and future projects?
19:45 : Public lecture : June 1914 and Philology for the 21st Century (Gregory Crane, Perseus Project, Tufts/Leipzig)
MOM, Amphithéâtre Benvéniste,
5/7 rue Raulin, 69007 Lyon
Wednesday, 4th June
Session 5 : Managing inaccurate quotations
9:00 Introduction. Definitions of the unstable notion of quotation. Brief survey on the quotation in the Antiquity (Smaranda Badilita, HiSoMA, Lyon)
Recent work on the concept of quotation in Antiquity ( see, for example C. Nicolas [ed], Hôs ephat’, dixerit quispiam, comme disait l’autre… Mécanismes de la mention et de la citation dans les langues de l’Antiquité, Recherches & Travaux, hors série, Université Stendhal, ELLUG, Grenoble, 2005 ; C. Darbo-Peschanski [ed], La citation dans l’Antiquité, Actes du colloque du PARSA Lyon, ENS LSH, 6-8 novembre 2002, coll. Horos, éd. Jérôme Millon, Grenoble 2004) underline the complexity of this « unstable » notion that affects several areas : the mode of transmission of texts (central issue in ancient studies); their reception (based much more on memory in Antiquity than today); the question of literary genres (some of which have the basic quote : the apophthegms the anthologies, the memorabilia); last but not least, social and ideological issues. The term « quotation » does not account for the complexity of this phenomenon in Antiquity, without being accompanied by a « tag cloud »: allusion, reference, narrations, Quellenforschung, paraphrase, rewrite, intertextuality. Some brief examples will illustrate this multiple shades table.
This paper discusses methodological issues regarding the analysis of new works from scratch in Biblindex, considering first the nature of a Father’s Bible and then describing the way of selecting and characterising a reference to this Bible in a patristic text. These include how to define the difference between allusion and quotation, how to distinguish and typify the Church Fathers’ ways of introducing and changing the biblical texts or how to decide where a quotation begins and where it ends. Through a few case studies, the solutions offered by Biblindex are presented and submitted for discussion.
Donatus’ Commentary on Terence contains many kinds of inserted texts: quotations from Terence’s comedies, and from other Latin and Greek writers, fragments of more or less unknown texts, truncated, abbreviated or inaccurate quotations, mentioned words or phrases. In this paper we intend to show how we have encoded these difficult passages according to TEI guidelines, and to discuss specific difficulties that we encounter now while preparing critical edition of the text.
10:05 La tradición literaria griega en los ss. III-IV d.C. gramáticos, rétores y sofistas como fuentes de la literatura greco-latina (Lucía Rodríguez-Noriega Guillén, University of Oviedo)
10:20 Dealing with all kinds of quotations (and their parallels) in a closed corpus: The methodology of the project “The literary tradition in the third and fourth centuries AD: Grammarians, rhetoricians and sophists as sources of Graeco-Roman literature” (Lucía Rodríguez-Noriega Guillén, University of Oviedo)
Our project aims to trace and classify all kinds of quotations (literal citations, paraphrases, lax references, imitations, parodies, mere mentions of authors and works), both explicit (with or without mention of the author and/or title) and hidden (imitations, parodies, allusions, use of material without mentioning its origin), in a corpus made up of the Greek grammarians, rhetoricians and “sophists” of the third and fourth centuries AD. At the same time, we try to detect whether or not these are first-hand quotations, and if our authors are, in turn, secondary sources of the same citations in later authors. We also study the philological (textual) aspects of the quotations in their context, and the problems of limits they sometimes pose. Finally, we are interested in the function of the quotation in the citing work. A coordinated project studies quotations in the Latin grammarians of the same period. In my talk I will briefly explain our methodology and how we store all those data in our file cards. In the future we would like to develop a dynamic web-site, but so far, and as a provisional solution, we use a simple desktop application called “Fichas”, which has been developed in C# language, under the .Net 4.0 Framework. The application allows the creation and modification of file cards (one for each quotation), which are stored individually in XML 1.0 format, and eventually exported to PDF format in order to publish them in the (static) website of the Project.
In this paper we want to raise the question of the relationship between on-going work on digital discovery of text re-use and the fact that the corpora of texts that are being searched are made up, at least in part, of narratives of various kinds. As Classicists working mainly from a traditional philological perspective, we focus on Greek and Latin epic poetry. This genre has the advantage of being so strongly codified in terms of its basic narrative features that it lends itself easily to a very simple type of formal narratological analysis. As a result, it also lends itself to an approach that brings into dialogue forms of literary intertextuality involving both specific kinds of text re-use and narrative features that operate on a non-linguistic level. Our question, therefore, is this: to what extent do we need to take into account original narrative contexts and structures when thinking about how to trace, analyse and represent digital analysis of text re-use?
11:25 Inaccurate citations as a source for the ECM of Acts – Misquotations or lost manuscript text? (Gunnar Büsch, INTF)
The presentation will deal with the methodology used for the ECM of Acts regarding patristic citations that contain text not extant in manuscripts. Special emphasis will be placed on the many differences of using such „inaccurate“ citations as textual witnesses for a critical edition of the New Testament, the main problem being the distinction between unintentional misquotations, deliberate changes by the citing author or even accurate citations of manuscript text unknown to us today. Examples from the ongoing work on the ECM will be used to show how and to what extent distinctions like this can be made.
11:50 QuotationFinder – Establishing the Degree to Which a Quotation or Allusion Matches Its Source (Luc Herren, London)
In the first paper on the QuotationFinder software, the tool and its purpose are introduced in a general way. This second paper is more technical and discusses how QuotationFinder produces an ordered list with the most exact quotation first and the loosest verbal parallel last. It is demonstrated how QuotationFinder produces a score for each potential quotation or allusion by assessing matched words along 5 parameters: quantity (the number of words matched within a reasonable number of lines); quality (whether the exact form of a word is matched, or a different form of the same word, or a form of a cognate); density (how close the matched words are to each other); rarity (how exclusive the matched words are); and, finally, order (to what degree the sequence of the words matches your search text.)
So far, QuotationFinder is aimed at researchers who are looking for quotations of and allusions to relatively small numbers of arbitrary, but short source texts in the vast collections of the TLG (Thesaurus Linguae Graecae) or the CLCLT (CETEDOC Library of Christian Latin Texts.) However, in this presentation we will be looking also at possibilities of extending QuotationFinder in order to facilitate searches of more and longer sources as well as searches in collections other than TLG and CLCLT and in languages other than Latin and Greek.
The Digital Athenaeus is producing the first digital edition of the Sophists at Banquet of Athenaeus of Naucratis. This work is a huge collection of quotations of lost authors and therefore a very interesting use case for implementing ontologies and data models for working with text reuses.
14:00 Towards a Cheatsheet for the Encoding of Biblical References and Quotations (Emmanuelle Morlock)
The aim of this workshop is to present the work done in the context of the Biblindex project and discuss a draft complementing the existing page on the wiki of www.tei-c.org (http://wiki.tei-c.org/index.php/Critical_Editions_Cheatsheet#Biblical_references_and_quotations).
14:30 Round-table Discussion : modelling a network of quotations, modelling a network of quotations. Sharing Encoding Patterns, Reference Systems and Concordances Tables.
Conclusions, opportunities for collaboration
15:30 Possible EU projects (H2020, 2. S6, COST, Culture Creative Europe) : Emilie Sablon (Lyon Ingéniérie Projet)
17:30 End of the meeting.
Link : http://calenda.org/287190